Diabetic retinopathy treatment depends on the stage of the disease, with early disease watched alongside diabetes control and vision-threatening disease treated with injections, laser, steroid implants, or surgery. Treatment protects vision but does not cure the underlying disease.
Diabetic retinopathy damages small blood vessels in the retina, the light-sensitive layer at the back of the eye. A national estimate for 2021 counted 9.60 million people in the United States living with diabetic retinopathy, 26.43% of people with diabetes. The same estimate counted 1.84 million with vision-threatening disease, 5.06% of people with diabetes.
Over years, high glucose damages the vessel walls, so fluid leaks or fragile new vessels grow. The condition is a leading cause of blindness among working-age adults in the United States. Control of glucose, blood pressure, and lipids slows its progression in adults with type 2 diabetes.
Which stages of diabetic retinopathy need treatment
Typically, advanced stages receive eye procedures, while early stages receive monitoring and diabetes control. Severe nonproliferative diabetic retinopathy (NPDR) marks high risk for progression toward proliferative diabetic retinopathy (PDR), while diabetic macular edema (DME) affects central vision.
| Retinopathy Stage or associated macular edema | What the 2026 ADA guidance and trials describe |
|---|---|
| No retinopathy | Eye exam every one to two years may be considered when glucose is at goal |
| Mild NPDR | No eye treatment; dilated exam at least once a year; glucose, blood pressure, and lipid control |
| Moderate or severe NPDR | Referral to an eye specialist; laser for some severe cases; preventive injections slowed progression but did not protect vision in one trial |
| PDR | Panretinal laser, or anti-VEGF injections as an alternative |
| Macular edema affecting central vision with vision loss | Anti-VEGF injections first-line; laser or steroids if swelling persists |
| Macular edema with good vision (20/25 or better) | Watching was as effective as treatment at two years in one trial |
Stages follow the international clinical scale published in 2003.
The American Diabetes Association advises referral to an eye specialist for any macular edema, moderate or worse NPDR, or any PDR. People with type 2 diabetes receive a first dilated exam at diagnosis, and people with any retinopathy receive a dilated exam at least once a year.
GLP-1 medicines and diabetic retinopathy
GLP-1 medicines do not treat retinopathy. The US labels for Ozempic injection and tirzepatide (Mounjaro, Zepbound) warn that a fast fall in glucose can briefly worsen existing retinopathy.
Tirzepatide works on two hormone receptors, glucose-dependent insulinotropic polypeptide (GIP) and GLP-1, and its labels ask that people with prior retinopathy be monitored, a warning listed with other tirzepatide side effects.
A study, SUSTAIN-6, compared semaglutide injection with placebo in 3,297 adults with type 2 diabetes and high heart risk over two years, with retinopathy complications in 3.0% compared with 1.8%. Among people with prior retinopathy, the rates were 8.2% compared with 5.2%, and the Ozempic label states that the long-term effect on retinopathy has not been studied.
A 2021 pooled analysis of six GLP-1 trials found no overall link, only higher risk with larger A1C drops, and the five-year FOCUS trial has no results until November 2027.
Blood sugar, blood pressure, and fenofibrate
Glucose, blood pressure, and lipid goals reduce the risk or slow progression of retinopathy in the 2026 ADA guidance. The Diabetes Control and Complications Trial compared intensive with standard glucose control in 1,441 people with type 1 diabetes over a mean of 6.5 years. Intensive control cut new retinopathy by 76% and slowed progression by 54% in people with mild disease at baseline.
In the United Kingdom Prospective Diabetes Study 38, tight blood pressure control slowed eye disease in people with type 2 diabetes and hypertension over nine years. Mean pressure was 144/82 mm Hg with tight control compared with 154/87 mm Hg with less tight control.
Tight control cut worsening of two or more steps on the retinopathy severity scale by 34%, and loss of three lines on an eye chart by 47%.
Fenofibrate, a triglyceride-lowering medicine, slowed retinopathy in three trials, yet the United States fenofibrate labels checked list only cholesterol and triglyceride uses.
The ACCORD Eye study compared intensive with standard glucose control, fenofibrate with placebo added to simvastatin, and intensive with standard pressure control in people with type 2 diabetes over four years. Progression was 7.3% compared with 10.4% for intensive glucose control and 6.5% compared with 10.2% for fenofibrate.
In the FIELD trial of people with type 2 diabetes, first laser treatment was needed in 3.4% on fenofibrate compared with 4.9% on placebo. The LENS trial compared fenofibrate 145 mg with placebo in 1,151 adults in Scotland with early retinopathy over a median of 4.0 years. Retinopathy that needed referral to an eye specialist, or treatment, occurred in 22.7% compared with 29.2%, and macular edema in 3.8% compared with 7.5%.
A further fenofibrate trial, Protocol AF, is still active, with primary completion expected in December 2029. In LENS, fenofibrate slowed retinopathy but did not change vision or quality of life, and kidney filtration was 7.9 mL/min/1.73 m² lower with it.
Anti-VEGF eye injections
Anti-vascular endothelial growth factor (VEGF) medicines block vessel growth and leakage after injection into the vitreous, the gel filling the center of the eye. The 2026 ADA guidance names anti-VEGF injections as first-line treatment for most eyes whose DME reaches the central part of the retina (center-involved DME) and impairs vision.
| Medicine | US approval for diabetic eye disease | Dosing schedule on the label |
|---|---|---|
| Eylea (aflibercept 2 mg) | DME and diabetic retinopathy | Every four weeks for five doses, then every eight weeks |
| Eylea HD (aflibercept 8 mg) | DME and diabetic retinopathy | Every four weeks for three doses, then every eight to 16 weeks for DME or eight to 12 weeks for retinopathy |
| Lucentis (ranibizumab) | DME and diabetic retinopathy | Once a month |
| Vabysmo (faricimab) | DME only | Every four weeks for at least four doses, then spaced by response, up to every 16 weeks |
| Beovu (brolucizumab)* | DME only | Every six weeks for five doses, then every eight to 12 weeks |
| Susvimo (ranibizumab refillable implant) | DME and diabetic retinopathy, after response to at least two injections | Refill every 24 weeks for DME or 36 weeks for retinopathy |
| Avastin (bevacizumab) | None for the eye; used off-label | Not on the label |
*Brolucizumab carries a warning for intraocular inflammation, retinal vasculitis, and retinal vascular occlusion, which may threaten vision.
Labels on DailyMed, checked October 2026; Pavblu and Cimerli biosimilars carry DME and retinopathy approvals.
Protocol T compared aflibercept 2 mg with bevacizumab 1.25 mg and ranibizumab 0.3 mg in 660 adults with center-involved DME and vision loss over two years. At one year the mean gains were 13.3 letters with aflibercept compared with 9.7 with bevacizumab and 11.2 with ranibizumab. Letters are read on a standard eye chart, where 15 letters is about three lines.
In that trial gains did not differ when starting vision was 20/32 to 20/40 but were larger with aflibercept when starting vision was 20/50 or worse. Protocol V compared aflibercept with laser and observation in 702 people with center-involved DME and vision of 20/25 or better over two years. Loss of at least five letters at two years occurred in 16% with aflibercept compared with 17% with laser and 19% with observation, with no significant difference.
Longer intervals reduce visits without smaller gains in two trials. YOSEMITE and RHINE compared faricimab with aflibercept in 1,891 people with DME over one year. Faricimab every eight weeks or spaced up to every 16 weeks was not worse than aflibercept every eight weeks, with gains near 11 letters.
PHOTON compared aflibercept 8 mg with 2 mg in 660 people with DME over 48 weeks. The 8 mg dose every 12 or 16 weeks after three monthly doses was not worse than 2 mg every eight weeks.
The labels state that serious procedure-related reactions occurred in less than 0.1% of injections and include infection inside the eye, retinal detachment, and pressure rise within 60 minutes. Susvimo carries a boxed warning of up to three times the rate of eye infection compared with monthly ranibizumab injections.
Laser treatment
Panretinal laser scars outer retinal areas to reduce new vessel growth in PDR. The Diabetic Retinopathy Study from 1976 established benefit, with severe vision loss in 6.4% of treated eyes compared with 15.9% of untreated eyes. Benefit was greatest where disc vessels or vitreous hemorrhage were present.
Focal or grid laser treats leaking areas in macular edema and is now second-line care. The Early Treatment Diabetic Retinopathy Study showed benefit for clinically significant macular edema. The ADA guidance now lists focal or grid laser as reasonable for swelling that persists despite anti-VEGF treatment.
Protocol S compared ranibizumab with panretinal laser in 305 people with PDR over five years. Vision gains were similar at two years and at five years, with mean vision near 20/25 in both groups, while field loss was worse with laser.
Vitrectomy was needed in 15% with laser compared with 4% with ranibizumab at two years, and new DME developed in 28% compared with 9%. The ranibizumab group received a mean of 19.2 injections over five years, so similar vision came with fewer complications but many more treatment visits.
Steroid implants for macular edema
Steroids reduce swelling for eyes with persistent DME despite anti-VEGF treatment or for eyes that cannot receive anti-VEGF treatment. Ozurdex is a dexamethasone 0.7 mg implant approved for DME. Retreatment occurs no sooner than six months, with a mean of 4.1 implants over three years in the MEAD trial.
Over three years in DME, the MEAD trial compared the implant with a sham procedure, one that imitates treatment without the medicine. A gain of at least 15 letters occurred in 22.2% compared with 12.0%.
Cataract developed in 68% of treated eyes that still have their natural lens compared with 21% with sham, and cataract surgery was needed in 61% compared with 8%. Eye pressure rise peaked near week eight, and 1% needed surgery for pressure.
Iluvien is a fluocinolone 0.19 mg implant lasting up to 36 months for DME in people previously treated with steroids without a large pressure rise. Most trial participants received one implant in three years.
A rise of at least 10 mm Hg occurred in 34% compared with 10%, and drops were needed in 38% compared with 14%. Pressure surgery was needed in 5% compared with 1%. Cataract developed in 82% of eyes that still have their natural lens compared with 50%, and surgery was needed in 80% compared with 27%.
Vitrectomy surgery
Vitrectomy removes blood-filled vitreous and treats detachments and scar tissue from PDR. Surgeons use the procedure for vitreous hemorrhage and for traction detachment.
Protocol AB compared aflibercept with vitrectomy plus laser in 205 eyes with vitreous hemorrhage from PDR over two years. Mean vision over 24 weeks was 59.3 letters with aflibercept compared with 63.0 letters with surgery, with faster recovery with surgery at four weeks. Both groups averaged near 20/40 at two years, and one third of eyes assigned to aflibercept later received vitrectomy.
Can treatment reverse diabetic retinopathy?
Treatment can preserve vision and may improve vision lost from macular edema or vitreous hemorrhage. However, permanent damage from retinal ischemia, scarring, or longstanding disease may limit recovery. Improvement in the retina seen on eye exams does not always mean better vision. PANORAMA compared aflibercept with sham in 402 eyes with moderately severe to severe NPDR without center-involved DME over 100 weeks. At 52 weeks, retinopathy improved by at least two severity-scale steps in 65.2% and 79.9% of eyes dosed every 16 and every eight weeks, compared with 15.0% with sham.
Vision-threatening complications or DME through week 100 occurred in 16.3% and 18.7% with aflibercept compared with 50.4% with sham. RISE and RIDE compared ranibizumab with sham injections in DME over 24 months, and retinopathy was less likely to worsen and more likely to improve with ranibizumab. A gain of at least 15 letters with the 0.3 mg dose occurred in 44.8% compared with 18.1% in RISE and 33.6% compared with 12.3% in RIDE.
Preventive injections slowed progression without protecting vision in early disease without edema. Protocol W compared aflibercept with sham in 328 people with moderate to severe NPDR without center-involved DME over four years.
PDR or vision-reducing DME at two years occurred in 16.3% compared with 43.5%. At four years the rates were 33.9% compared with 56.9%, while vision change was minus 2.7 letters compared with minus 2.4 letters. The injections slowed progression to sight-threatening disease, but vision at four years was no better than with sham.
How often treatment is needed and what it costs
Treatment continues for years with regular exams and repeat procedures. Most people starting ranibizumab or aflibercept need injections every four to eight weeks during the first 12 months, with fewer later. Faricimab and aflibercept 8 mg can extend to every 16 weeks in DME.
The ADA guidance notes that anti-VEGF treatment for PDR needs more visits than laser, and that lapses in follow-up risk worse outcomes. A five-year cost analysis of Protocol S found four times the injections and three times the visits with ranibizumab compared with laser. In Protocol T the median was 15 to 16 injections over two years, with five to six in the second year.
Focal or grid laser was added in 41% to 64% of eyes across the three Protocol T groups. No eye drop is approved in the United States for diabetic retinopathy or DME among the labels checked.
Current diabetic retinopathy treatment cost per injection, laser session, implant, or surgery was not published in the sources checked. A 2016 analysis found aflibercept and ranibizumab about 31 and 20 times more costly than bevacizumab using 2015 to 2016 United States prices.
Note: This page offers general information and not medical advice. Seek urgent eye assessment for sudden vision loss, new flashes or numerous floaters, a curtain-like shadow, or increasing pain/redness after an injection.
Sources
- US prevalence, staging, and guidance: Wilkinson et al., 2003; Lundeen et al., 2023; American Diabetes Association, Standards of Care section 12, 2026; National Eye Institute, Diabetic retinopathy, 2025
- DRCR Retina Network trials: Protocol T, 2015; Protocol S, 2015, and five-year results, 2018; Protocol V, 2019; Protocol AB, 2020; Protocol W, 2021, and four-year results, 2023
- Other eye trials: RISE and RIDE, 2012; MEAD, 2014; PANORAMA, 2021; YOSEMITE and RHINE, 2022; PHOTON, 2024
- Glucose, blood pressure, and fenofibrate trials: DCCT, 1993; UKPDS 38, 1998; FIELD, 2007; ACCORD Eye, 2010; LENS, 2024
- GLP-1 evidence: SUSTAIN-6, 2016; Bethel et al., 2021
- US prescribing labels on DailyMed: Lucentis, Beovu, Susvimo, 2025; Eylea, Eylea HD, Vabysmo, Ozurdex, Iluvien, Ozempic injection, Mounjaro, Zepbound, 2026
- Cost analyses: Ross et al., 2016; Hutton et al., 2019